SoftZoo is a cutting-edge platform developed by a team of MIT researchers that allows engineers to explore and optimize designs for soft robots. Soft robotics is a field that focuses on building flexible machines inspired by the movement of animals in the natural world. SoftZoo takes this concept further by providing a bio-inspired platform for studying co-design in soft robots.
The platform features 3D models of animals like pandas, fishes, sharks, and caterpillars, which can be used to simulate various tasks such as locomotion, agile turning, and path following in different environments. SoftZoo allows users to assess the performance trade-offs of different designs in different terrains, including snow, desert, clay, and water.
What sets SoftZoo apart from similar platforms is its ability to model movement that reacts to the physical features of different environments. This is made possible by a differentiable multiphysics engine, which enables the simulation of multiple aspects of a physical system simultaneously. The engine’s differentiability optimizes co-design by reducing the need for costly simulations, allowing users to design and control soft robots with more sophisticated algorithms.
SoftZoo highlights the importance of morphology, the study of the shapes, sizes, and forms of organisms, in designing effective robots. Depending on the environment, certain biological structures are more optimal than others. By drawing inspiration from nature, researchers can develop specialized artificial lifeforms tailored to specific terrains. For example, the undulating geometry of a jellyfish can inspire the design of efficient water-traversing robots, while the unique wing structures of dragonflies can inform the development of agile flying robots.
Traditionally, robots faced challenges navigating cluttered environments because their bodies were not compliant with their surroundings. However, SoftZoo enables designers to co-optimize the robot’s brain and body simultaneously, resulting in robots that are more aware and specialized for their tasks. These advancements in behavioral and morphological intelligence make soft robots more capable in rescue missions and exploration, such as navigating floodwaters to find missing individuals.
SoftZoo provides an open-source simulation platform for soft robot designers, making it easier and more flexible to build real-world robots with enhanced locomotion capabilities in diverse environments. By co-designing the body and control algorithms of robots, SoftZoo allows for the rapid creation of customized machines for specific tasks.
While SoftZoo is currently limited to simulation, the researchers acknowledge the challenges of transferring designs to physical robots using current fabrication techniques. They are actively working on overcoming these challenges, which involve muscle modeling, spatially varying stiffness, and sensorization.
Looking ahead, the creators of SoftZoo envision applications in human mechanics, particularly in the area of manipulation. By simulating more human-like tasks, such as grasping, moving, and stacking objects, the platform can assist soft robotics designers in assessing the performance of soft robotic arms and other manipulative systems.